This is not so much the Framework issue but just Linux in general
Here is what is happening:
I have Fedora 39 running on my system
I have a VM of Windows 11 running on either Virtual Box and VMWare Workstation (testing both, not at the same time)
in both situations VPN connections will drop after 10-20 mins regardless of the hypervisor software i test
VPN on the base metal host OS runs flawlessly (have tested till the 24hour timeout kicks me)
so i KNOW this is a virtualization issue
I’ve tested this in Windows and this doesnt reproduce, so its something im doing with Linux host and Windows VM (i have NOT tested Linux VM - thats not a useable config for my needs)
Has anyone any thought as to where i can start looking to see why this happens?
This is a tricky one as I have not run a VPN inside of a VM in recent memory myself. Setting aside the VM nature, this does feel like kill switch behavior. So it may make sense to triple check the settings/behavior in the Windows VM with this in mind.
Why type of connection is the vm using at its base? Is it a NAT, Bridge to LAN with a static ip, MACVLAN, Direct Attachment etc.? Is the connection “idle” when it drops i.e. no discernible traffic for a period of time or is it getting punded and then drops, or something in between? What VPN is it being used with? NordVPN or similar provider, an OpenVPN to your home network, a site to site VPN, an appliance based VPN? Which one is the priority for you VMWare or Virtual Box or are you trying to settle on a virtualization solution which if that is the case why not KVM since it is built into the kernel?
Why type of connection is the vm using at its base? Have tried NAT and Bridged - same behavior
Is the connection “idle” when it drops i.e. no discernible traffic for a period of time or is it getting punded and then drops, or something in between? It will drop sooner if traffic is heavier, usually
What VPN is it being used with? Fortinet VPN licensed corperate to my office at work
Which one is the priority for you VMWare or Virtual Box or are you trying to settle on a virtualization solution which if that is the case why not KVM since it is built into the kernel? VirtualBox runs better, so i’d stick wit hthat if possible (VMWare has its own bugs right now and personally dont want to fight those demons)
No worries. Based on what you shared, I’d be interested in seeing the behavior happening again, then on the affected environment, run rfkill and see if wlan or similar has been soft blocked.
You mention in a reply that you are using Fortinet VPN. I’m not familiar with that. How have you configured your VPN? Are you using the Fortinet VPN client? Are you using something standard provided by Fedora? If so is it supported by Fortinet or your company? Have you reached out to either of them for support?
im using the fortinet official forticlient
licensed through work
I am the company support LOL
I have been in touch with fortinet support too and we are kind of at a wall as the behavior seems to be so odd
Are you able to capture debug logs from fortinet, to try to catch why the connection is dropping? Do you see the same behavior in a kvm vm (presumably yes, but worth a try). Finally, are you able to catch anything with a tool like wireguard - to be clear, I don’t know if it would with a VPN connection, but also may be worth trying.
I do have teh Fortilogs
The error is 2747 on FC7.2.2
WSAENOBUFS
10055
No buffer space available.
An operation on a socket could not be performed because the system lacked sufficient buffer space or because a queue was full.
I can recreate the issue in both Virtualbox AND VMWare Workstation.
i’ve no experience with KVM as a workstation host - i an read up and see about testing that scenario
